MIT OpenCourseWare
http://ocw.mit.edu
HST.583 Functional Magnetic Resonance Imaging: Data Acquisition and Analysis
Fall 2006
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
HST.583: Functional Magnetic Resonance Imaging: Data Acquisition and Analysis, Fall 2008
Harvard-MIT Division of Health Sciences and Technology
Course Director: Dr. Randy Gollub.
Final Exam (and answer key)
HST.583
December 15, 2008
Instructions:
There are 7 problems. Points for each problem are indicated. Please immediately
confirm you have all X pages! This is a closed book/ closed notes exam. You have two
hours.
Read the whole exam first. Write legibly. Describe what you are thinking. Even if the
final answer is wrong, you will be given credit for your reasoning and approach to the
problem. Be efficient, assess what you can answer right away; divide your time so that
you can cover all the problems, allocating less time to the easiest parts. Don't spend too
much time on a single question.
Problem 1:
a) List the major brain regions that comprise the “Default mode network” as originally
described by Marcus Raichle and colleagues in 2001 (3 points)
b) describe the hypothesized role of decreased activation in this network during attention
demanding tasks from a cognitive neuroscience perspective (3 points)
c) give an example of how, assuming that this hypothesis is true, this information could
be used to improve the sensitivity or specificity of an fMRI experimental paradigm (3
points).
[Answer:
a) Extrastriate visual areas, posterior cingulate, precuneus, medial prefrontal cortex, (+/orbital frontal cortex)
b) decrease in activation when shifting from rest to attention demanding tasks (stimulusindependent, task-independent)
c) Many answers will be accepted. Here are examples of good ones.
-Since the differences between rest with eyes open and closed is still
controversial, a paradigm that ensures that during the baseline condition all subjects
maintain the same state (eyes open and fixated on center of screen or closed) will
decrease between subject variance and perhaps even improve within subject
significance of results.
-Any method to ensure that during “baseline” subjects are indeed resting quietly
with eyes closed or open and not doing another “off paradigm” task (recalling list of
errands to run after experiment is over)- physiological monitoring, low level task- fixate
cross hair, etc. will give more consistent results within and across subject.]
Problem 2:
A physics post-doctoral fellow in Dr. Adelsteinsson’s lab shows you a generalized pulse
sequence for a common imaging technique used boost T1 contrast. The pulse
sequence begins with a 180 degree inversion pulse, and is followed by a delay (TI).
After the delay imaging commences:
Figure 1.
a) This technique is known as __inversion___ __recovery___, and was used in Dr.
Triantafyllou’s lab to measure T1 from different tissues. (2 points)
Because the 180 degree pulse flips the net magnetization to the negative state, it
effectively doubles the dynamic range of the signal. The Mz magetization profile for
white matter at 3.0 T looks like this:
Figure 2.
The physicist points out to you that it is possible to eliminate MR signal from a particular
tissue type by properly choosing TI.
b) Qualitatively show on the above figure the TI that would eliminate white matter signal.
Explain your choice. (3 points)
Draw at zero crossing. Mz will be zero here, so when you excite with your 90 for
imaging, no signal will be knocked down into the transverse plane (Mxy) and no signal
will be detected by the receiver coils.
c) Why would it be useful to eliminate white matter signal when performing functional
imaging analyses? (1 point)
For example if you want to segment out gray matter exclusively - this gives you excellent
white matter/gray matter contrast (the best possible actually, since white matter is now
dark!).
d) The T1 of white matter is 832 ms. Compute the TI needed to null white matter signal.
(Hint: Modify the longitudinal (T1) magnetization equation to accommodate double the
dynamic range). Points for original equation, modified equation, and solution. (3 pts)
Longitudinal relaxation
Mz = Mo(1-e-t/T1)
IR equation
Mz = Mo(1-2e-TI/T1), for double dynamic range.
Now set Mz to zero and solve for T1; should be around 576.7 ms.
Problem 3:
Given a stimulus schedule where Condition 1 is presented at t=2 and t=14 sec and
Condition 2 is presented once at t=6sec
a. Write out the FIR design matrix assuming TR=2s, Number of time points =10, and the
FIR is window=8s. Include regressors for offset and linear drift (5 points).
Answer:
0 0 0 0 0 0 0 0 1 1
1 0 0 0 0 0 0 0 1 2
0 1 0 0 0 0 0 0 1 3
0 0 1 0 1 0 0 0 1 4
0 0 0 1 0 1 0 0 1 5
0 0 0 0 0 0 1 0 1 6
0 0 0 0 0 0 0 1 1 7
1 0 0 0 0 0 0 0 1 8
0 1 0 0 0 0 0 0 1 9
0 0 1 0 0 0 0 0 1 10
Note: the offset and trend could be the 1st two columns (this affects the contrast matrix
below).
b. Write out a contrast matrix that tests for the difference between condition 1 and
condition 2 at 2s post-stimulus (5 points).
[0100
0 -1 0 0
00]
What are the trade-offs between using an FIR model and assuming a shape to the HRF?
(5 points)
[ Advantages of the FIR: guaranteed to fit the data the best and can handle in HRF
shape. Advantages of assuming a shape: increased statistical efficiency and a higher
number of DOFs.]
Problem 4:
One controversial claim for BOLD fMRI is that the positive change in signal amplitude is
preceded by a smaller decrease in signal. This postulated decrease has been termed
the "initial dip" or “early dip”:
Source: Duong et al. MRM 44 (2000): 231-242.
Courtesy of Wiley-Liss, Inc., a subsidiary of John Wiley & Sons, Inc. Used with permission.
a) Suggest a possible physiological mechanism for the initial dip. (5 points)
b) If the initial dip does exist, why could it be useful for functional imaging? (5 points)
c) Why do you think it has been difficult to observe with fMRI? (5 points)
[Answer:
a) The initial dip may represent the initial increase in CMRO2 before blood flow
increases to deliver more oxygen (see Malonek and Grinvald, Menon). As a result, the
BOLD signal will initially go down, since there is transiently more deoxyhemoglobin in
the local blood, since oxygen is being removed and metabolized by the neuron. As CBF
starts to increase, delivery of oxygenated blood will flush out the dHb, leading to the
main positive response.
Also acceptable: the Buxton et. al. balloon model can also explain the initial dip, based
on a non-linear flow out/ volume relationship. That is, if the curve initially has a shallow
slope, so that with a sharp increase of Fin the first effect is a rapid filling of the balloon
(because Fout is slow to increase), then the dHb content initially would increase due to
the increased venous volume. This would also result in an initial dip.
b) Because CMRO2 should be more spatially specific to neural activity (since an increase
in CMRO2 is closely associated with firing neurons) the initial dip may be a more
spatially accurate measure of activation. Some believe that CBF may not be regulated
with as fine a spatial control as CMRO2 (recall “watering garden for sake of thirsty
flower”).
c) Much smaller signal, thus harder to resolve given the limits imposed by the poor
contrast to noise of fMRI BOLD signals.]
Problem 5:
List three possible ways CBF may be regulated in the brain; give a brief description of
each. (5 points)
[Answer:
It is increased by an increase in capillary flow, not capillary recruitment. Capillary flow in
the brain is regulated by
1- modulation of arteriolar diameter by vasoactive substances (NO, CO2, K+, adenosine)
that bind to vascular smooth muscle receptors (glia, neurons, ?)
2- secretion of vasoactive substances (Adenosine, NO, Prostaglandins, H+, K+) by
neurons that are the byproducts of energy metabolism into surrounding neuropil
3- direct neural innervation of arterioles (NO, VIP, DA, Substance P, NA, GABA, 5-HT,
ACh)
4- indirect control by substances (K+, Adenosine, CO2) released via astrocyte endfeet
5- Pericyte constriction at capillary level
Problem 6:
a) When registering the fMRI data from one of your subjects to his/her own structural
MRI data collected in the same scan session, what are the sources of noise (variance)?
For each source of variance listed, indicate what you could do to minimize the variance
(5 points).
b) Describe one method for inter-subject registration of the fMRI data from the subjects
in one of your experiments so that you can perform a group analysis. What are the
sources of noise (variance, error) inherent in this method? (5 points)
[Answer:
a) (at least 2 of these)
--- scanner orientation- ensure subjects are placed in scanner in the same “sweet spot”
of scanner and that acquisition parameters are matched for the anatomical and
functional scans- e.g. slice orientation
--- image resolution- can’t eliminate, optimize to obtain best possible resolution of both
scans
--- EPI distortions/imaging physics- low bandwidth of scan acquisition, shim before
acquisition, good scanner maintenance, collect field maps and perform a postprocessing distortion correction, make sure brain is centered in “sweet spot of scanner”
--- normal range of anatomical variation in subjects- nothing can change this but
improved registration algorithms can decrease the negative impact on the results of the
analyses
--- pathological abnormality in anatomy- screen subjects, have exclusion criteria to drop
subjects if abnormalities are identified on their scans.
b) All the data must be registered to a common “space” the registration can be either
landmark based (e.g. average of the cohort or single subject of the cohort, Talariach,
MNI305) or image features (e.g. intensity based). Either of these methods can use
many different warping algorithms (e.g. affine, spline, cortical surface). Sources of noise
include inter-subject variability in anatomy, inconsistency in image acquisition
parameters (e.g. cross site or after upgrade), poor head positioning in scanner with large
differences in distortions]
Problem 7:
One of your best friends aspires to do an fMRI experiment to test their hypothesis about
how the brain shifts attention from doing a single task to doing two simultaneous tasks.
The hypothesis is that there is only a “fixed” amount of attentional ability in the brain and
that when shifting from one to two tasks the ability will be “divided” and hence less will
be available for either one of the two tasks. The trouble is that they know nothing about
fMRI. Since you have just completed HST.583, your friend assumes you are an expert
on this topic and has come to you for advice. Most importantly, there is a rich donor who
has given unrestricted funds to pay for the scans, subject remuneration and generous
consulting fees to you, the expert, for providing the following information. So you accept
the job (just in time to pay for your Holiday fun!) and get to work.
(Hint: No knowledge of the cognitive neuroscience in this domain beyond what you
learned in the course is required; just apply your own best logical effort and common
sense to that base. If you happen to know specific, pertinent references please cite and
use them in your response. Take your time thinking all the way through this so that you
do the best job you can. Five bonus points are awarded for a fully integrated response.)
Answer:
Each answer to this problem is unique and is graded by the Course Director for clarity,
accuracy, validity and feasibility.
Design an fMRI experiment to test the hypothesis. Provide:
i. A description of the putative functional neuroanatomic attentional network you will be
investigating. (5 points)
i. Might include (top- down) parieto-frontal (eye fields), cingulate and ascending reticular
activating system (bottom-up) network for spatial attention, occipitotemporal network for
object/face recognition, prefrontal network for attention & comportment,
ii. A simple statement of the hypothesis you are testing. Include an explicit statement
of what and where you will be measuring to determine if H0 is true or false. If you are
able to give an estimate of an appropriate Power analysis please do so, and/or specify
the number of subjects that will be required to test the hypothesis. (5 points)
All answers must include the following: fMRI data is noisy, signals are small and
subjects highly variable. With too small of a sample size, you may miss a real finding
completely if the activations in regions are near threshold for detection or if have much
inter-subject variability in localization. This is especially important for the smaller brain
regions. A minimal sample size to draw meaningful conclusions is typically 12-16
(anecdotally, for many fMRI studies). The actual required sample size, however,
required depends on several factors; specifically the expected effect size and the desired
power.
iii. A description of the fMRI paradigm that includes exactly what the subjects will be
doing and when and how. (5 points)
An example of a correct response is:
First experiments typically begin with a block design to maximize power to detect
regional activation. She is proposing to spend 6 10 minute long scans doing an event
related paradigm. A less risky proposal would be to invest in a few block design trials to
identify the voxels of interest and then use the event related paradigm to look at the
hemodynamic response times of the components of the network. Also, she made no
mention of parametrically varying the motor task to allow a more robust detection of
movement correlated activation. Such variations could include the dimension of time
(duration of movement), complexity, numbers of fingers used.
A minimum of two event types must be specified; complex finger movement and imagined
finger movement and the duration of each event specified. Adequate time must be given
for rest (control condition). Each trial should be 1-5 sec long. We need plenty of trials to
have statistical power, so given 10 minutes scan could do about 100 trials per scan.
Either spaced events spaced about 16 sec or rapid presentation with 5-10sec with jitter.
The number of trials needs to be provided along with information about whether the
events are being timed for a block, spaced single trial or rapid event related analysis.
To be able to average data across all scans, need to have stable performance of the task
during real movement so must practice prior to scans until achieve plateau performance.
This also indicates need to collect behavioral data to verify performance accuracy, speed
etc. Otherwise, learning/memory effects may confound results. This training will also
help to equilibrate the attention level of subjects and will help equilibrate the imagery
component across subjects.
Keeping still for 10 minutes is difficult for some subjects; a larger number of shorter
scans (~5minutes) may yield better quality data with less motion artifact.
iv. A description of the experimental set-up as needed to perform the experiment. (5
points)
As noted above, each answer is unique and is graded by the Course Director for clarity,
accuracy, validity and feasibility as well as how well it will enable the proposed design.
v. A list of the image acquisition sequences you will collect including key parameters
and a justification for each (why you need it, what you will use it for). (5 points)
anatomical image
What matters is good contrast between gray and white matter. Use a sequence that has
high spatial resolution and T1 contrast. PD does not. They should recommend an
MPRAGE or FLASH or SPGR type scan with the highest spatial resolution (smallest
voxels) available.
fMRI scans
The strongest BOLD contrast is obtained with T2* weighted images, like gradientecho EPI. Spin-echo EPI offers less susceptibility distortions but 5 times less BOLD
sensitivity. So, should start using the imaging method that is most sensitive. Unless
of course the regions of interest for activation are in or close to an area where there
are strong susceptibilities, then should try spin-echo EPI. But I would start with
gradient-echo EPI. Spiral is okay too if that is what is available.
vi. A detailed description of your proposed data analysis methods, including any preprocessing steps, for the single subject and group analyses. (10 points)
All answers must include not only a correctly described sequence of processing steps but
also a justification for each and reasonable parameters/settings where indicated.
Minimal processing includes registration to atlas space, skull stripping, motion
correction, (+/- artifact detection), spatial smoothing, model fitting with statistical
threshold parameters.
vii. Your prediction of what you will find when the study is completed. (5 points)
Each answer is unique and is graded by the Course Director for clarity, validity with
respect to the proposed design and analysis methods and feasibility.